Python requests

09 Feb 2018 on Computer Science

from bs4 import BeautifulSoup
import requests
import urllib.request

df = pd.DataFrame(columns=["rank","keyword"])
response = requests.get("https://www.officeholidays.com/countries/usa/2016.php")
bs = BeautifulSoup(response.content, "html.parser")
trs = bs.select("table td")
trs1 = trs[1::5]

li = []
holi = pd.DataFrame()
count = 0

for i in trs1[0:14]:
    li.append((i.text).strip())
    li[count] = '2016 ' + li[count]
    li[count] = li[count].split(" ")
    li[count] = li[count][0] + "-" + li[count][1] + '-' + li[count][2]
    count += 1

holi['date'] = li
holi['date'] = pd.to_datetime(holi['date'])
holi

	date
0	2016-01-01
1	2016-01-18
2	2016-02-15
3	2016-04-15
4	2016-05-08
5	2016-05-30
6	2016-06-19
7	2016-07-04
8	2016-09-05
9	2016-10-10
10	2016-11-11
11	2016-11-24
12	2016-11-25
13	2016-12-26

---

Reference